With the increasing role of the stock market and the diversity of its instruments, the main issue has been an increase in volatility, and the question of predicting the volatility of global stock markets is becoming increasingly important. In this paper, four modifications of the ARCH / GARCH volatility model are considered. The main task was to identify the model closest to the real market situation.The problem of choosing models is that each of them takes into account some properties of the time series. Using historical data, time series were modeled and all the above models were evaluated. Student’s criterion was used to compare different volatility models. As a result of the study, the following results were obtained: the volatility obtained from the IGARCH model is more useful for understanding market sentiment than other models. In addition, this model takes into account some of the unique properties of the time series and does not care about the long-term dispersion, which explains the market behaviour